Automatic Redundant Routing in Exchange 2007 (Full Version)

All Forums >> [Microsoft Exchange 2007] >> Message Routing


ponzekap2 -> Automatic Redundant Routing in Exchange 2007 (5.Oct.2008 7:56:51 PM)

At my company we have two AD sites, and I have configured our Exchange 2003 routing topology to mimic this setup.  We have also two non-exchange SMTP gateway servers at each site, that all outgoing email flows out of.  Our current setup is we have SMTP connectors in each routing group, that have an address space of * and a cost of 5, that route all mail through their respective gateway servers, in their site. 

To create a redundant enviornment, I have also created another SMTP connector in each routing group, with a cost of 7, that routes email from Site A, out through the gateway server in site B, in case of a server failure in Site A, or loss of internet connection.  This allows outgoing email to automatically start flowing out to remote site, so that there is virtually no downtime for outgoing emails.  This works great for us.

I have begun designing our Exchange 2007 migration, and in my tests, it seems that the same setup with multiple send connectors does not work in Exchange 2007.  From reading Microsofts documentation, it seems that this is by design.  Exchange 2007 does not use link status updates, so it will not re-calculate a route if a server is down, so essentially admin interference is neccessary.  I know that I can create a send connector, and specify both of the gateway servers as smart hosts, and this will work, but I also dont want this to be used in a round robin approach.

What is the best way, and what have you guys used, to create the above enviornment in Exchange 2007.  Essentially each site will have one hub transport and one mailbox role installed.  Each site will have its own dedicated non-exchange smtp gateway server.  If the gateway server in site A fails, how do i get email to route out through the gateway server in site B, besides adding this as a second smart host in the send connector?

Sorry for the lenghty post.

josh -> RE: Automatic Redundant Routing in Exchange 2007 (6.Oct.2008 8:12:52 PM)

ponzekap2 -> RE: Automatic Redundant Routing in Exchange 2007 (6.Oct.2008 8:16:01 PM)


Thanks for the reply, but thats not what I want.  I am not worried about NLB the hub transport servers, but instead want to configure redundant routing topology for External email.

So if the gateway email server in NY site goes down, external email will flow from NY out to the London gateway server.

Elan Shudnow -> RE: Automatic Redundant Routing in Exchange 2007 (6.Oct.2008 8:22:07 PM)

Something like the following?
Set-SendConnector "Internet Connector to local SMTP Appliance" -AddressSpaces "*;1"
Set-SendConnector "Internet Connector to other site's SMTP Appliance" -AddressSpaces "*;2"

ponzekap2 -> RE: Automatic Redundant Routing in Exchange 2007 (6.Oct.2008 8:24:27 PM)

Yea, but I've tried this.  If the primary connector fails, like the remote smart host is down, it never fails over to the second connector. 

Elan Shudnow -> RE: Automatic Redundant Routing in Exchange 2007 (6.Oct.2008 8:29:43 PM)

Hm, should work.  I'm assuming you have SP1 and Rollup 3 for SP1 installed?  Have you tried turning up diagnostic logging (Set-EventLogLevel) and maybe it'll give some clues as to why it's not failing over to the next connector?  Or maybe run BPA and it'll elude as to some misconfiguration?

If none of the above really work, it really sounds like a bug otherwise there'd be no point to those costs.  A call to PSS probably would be needed.

ponzekap2 -> RE: Automatic Redundant Routing in Exchange 2007 (7.Oct.2008 12:04:45 AM)

See, this is the documentation I was referring to.  It seems that this specific type of situation wont allow the message to be re-calculated.

Message delivery failures that are caused by network connectivity problems are not detected by the routing component of the categorizer. When an SMTP connection cannot be established to the next hop target, SMTP Send retries the queue. The MaxIdleTimeBeforeResubmit parameter, which is located in the EdgeTransport.exe.config file, has a default value of 12 hours. After the configurable retry interval (MaxIdleTimeBeforeResubmit) has expired without successfully establishing a connection, all messages from the delivery queue are resubmitted to the Submission queue. If the connectivity problem still exists, this process is repeated. If the connectivity problem is resolved, messages are delivered as soon as message retry is successful. Or, a configuration change that modifies the next hop destination may resolve the problem. For example, if the problem is caused by all the Hub Transport servers in a destination site being offline, and you move mailboxes to a server in a different site, the next hop would change to the new site.
[image],EXCHG.80%29.gif[/image]Note: The auto resubmission from the message delivery queue to the Submission queue only happens for non-connector queues. Connector queues remain in retry mode until the problem is fixed, or the messages expire and a non-delivery report (NDR) is sent.

Exchange_Geek -> RE: Automatic Redundant Routing in Exchange 2007 (7.Oct.2008 2:42:10 AM)

All this sums up to one line - (correct me if i am wrong) Exchange 2007 has removed the concept of link-state update technology which was present in Exchange 2003.

This means that if a connector has been specified to send emails - Exchange WOULD NOT try to send email over any other connector if the specified connector goes down.

We had a concept to re-categorize messages even if messages were stuck for a particular queue in Exchange 2003. This i believe has been removed in Exchange 2007.

So, if we are expecting the redundancy in mail flow by creating second connector for the same RG - this wont work.

ponzekap2 -> RE: Automatic Redundant Routing in Exchange 2007 (7.Oct.2008 9:39:35 AM)

Exactly, which is what I said previously. 

you can always specify the smart host in the same connector, but then you get load balance, which may not be ideal.

Page: [1]